Cumulative absolute velocity (CAV) is an important ground motion intensity measure used in seismic hazard analysis. Based on the Next Generation Attenuation strong motion database, a simple ground-motion prediction equation is proposed for the geometric mean of as-recorded horizontal components of CAVs using mixed regression analysis. The proposed model employs only four parameters and has a simple functional form. Validation tests are conducted to compare the proposed model with the recently developed Campbell–Bozorgnia (CB10) model using subsets of the strong motion database, as well as several recent earthquakes that are not used in developing the model. It is found that the predictive capability of the proposed model is comparable with the CB10 model, which employs a complex functional form and more parameters. The study also corroborates previous findings that CAV has higher predictability than other intensity measures such as the peak ground acceleration. The high predictability of CAV warrants the use of the proposed simple model as an alternative in seismic hazard analysis. Copyright © 2012 John Wiley & Sons, Ltd.